Artificial intelligence in medical practice is the use of computer techniques to perform clinical diagnoses and suggest treatments in medical areas [1]. It has the capability of detecting meaningful relationships in a data set and could be used for the diagnosis, treatment, and for coming to a particular conclusion. Similar to the way doctors are educated through years of medical schooling and learning from mistakes, artificial intelligence algorithms learn to do the same job as a doctor. They perform tasks requiring human intelligence like pattern and speech recognition, image analysis, and decision making. The Artificial Intelligence algorithm includes feeding data in the computer system, which are structured having a label recognizable to the algorithm, performance is analyzed just like exams give an analysis of a medical student’s performance thus giving results. Based on the results of this analysis the algorithm can be modified, fed more data, or rolled out for the decision-making of the person writing the algorithm [2].
Fig: AI algorithm learning the basic anatomy of a hand and can recreate where a missing digit should be. This could allow physicians to see the proper place to reconstruct a limb/put a prosthetic.
These performances and results are tested with a physician’s performance to determine its clinical ability and value. In medicine language, it includes input data based upon numericals such as Heart Rate or Blood Pressure and based upon images such as Magnetic Resonance Imaging Scans or Images of Biopsy Tissue Samples. The algorithms from this data could be a probability or a classification. The result of the above example could be the probability of having an arterial clot according to the heart rate and blood pressure data or the labeling of an imaged tissue sample by classifying it as cancerous or non-cancerous. There are two recent applications in the Artificial Intelligence of clinical and accurate algorithms benefiting both patient and doctor for the diagnosis. One is the algorithm researchers at Seoul National University Hospital and College of Medicine developed called Deep Learning-based Automatic Detection to analyze chest radiographs and detect abnormal cell growth (cancers). The results were compared to many physician’s detection abilities and were found to perform better than the doctors [2].
Fig: Artificial Intelligence Algorithm. Left panel showing the image fed into an algorithm. The right panel shows a region of potentially dangerous cells, as identified by an algorithm, that a physician should look at more closely.
Fig: Artificial Intelligence algorithm; Deep Learning Method.The left panel shows the original X-ray. The right panel shows the X-ray with orange color indicating signs of pneumothorax which could be unnoticed by radiologists
The second algorithm was developed by researchers at Google AI Healthcare called Lymph Node Assistant which analyzed histology slides stained tissue samples to identify metastatic breast cancer tumors from lymph node biopsies. It could identify suspicious regions of the sample which could not be distinguished with the human eye. It was proven to accurately classify a cancer as cancerous or non-cancerous in 99% of the cases. Hence these algorithms could help doctors with correct diagnosis thus allowing them to invest time in solving cases that computers cannot solve [2].
Fig: AI algorithm; Lymph node biopsy
Artificial intelligence could be considered a boon as it may help for early diagnosis of diseases whose later diagnosis can cause delays in the treatment and may be harmful to the patient. For example, researchers have claimed that it could be used to diagnose Alzheimer’s disease years before symptoms appear. The computers can be trained for brain scans to be able to spot subtle signs of dementia that could be missed by humans allowing early diagnosis. This could probably be done using 18-F-fluorodeoxyglucose positron emission tomography (FDG-PET). In an FDG-PET scan FDG, a radioactive glucose compound is injected into the blood. PET scans can then measure the uptake of FDG in brain cells, an indicator of metabolic activity. Through Deep Learning, the algorithm can teach itself metabolic patterns that correspond to Alzheimer’s disease.If one can detect the symptoms earlier, it would help investigators to find better ways to reduce or halt the disease process. Future research should take into consideration, training the deep learning algorithm to look for patterns associated with the accumulation of beta-amyloid and tau proteins, abnormal protein clumps, and tangles in the brain that are markers specific to Alzheimer’s disease, according to UCSF’s Youngho Seo, Ph.D., which can add another dimension to using Artificial Intelligence in Alzheimer’s disease detection [3].
Fig: Fluorine 18 fluorodeoxyglucose PET images from Alzheimer’s Disease Neuroimaging Initiative set preprocessed with the grid method for Alzheimer disease patient
Artificial Intelligence has many clinical applications to improve patient care and potentially save lives. Maintaining medical records and past history is the first step in health care where robots collect, store, reformat, and trace data to provide faster and more consistent access. They also analyze data including notes and reports from a patient’s file and clinical expertise to help to choose the right treatment pathway [5].There are some latest tools and technology developed in the health care sector based on the Artificial Intelligence algorithm. This includes: MelaFind, which is a tool that does not involve introduction of instruments into the body and gives extra information to dermatologists in early detection and recognition of skin cancer,lesions and helps in it’s examination. It also helps in evaluation of skin lesions up to 2.5 mm beneath the skin. By using Artificial Intelligence based algorithms, dermatologists can analyze irregular moles and diagnose serious skin cancers such as melanoma. The device demonstrated 98.3% sensitivity by correctly identifying 172 out of 175 melanomas and high-grade lesions. Robotic-assisted therapy is used in neurological patients and is specially used for stroke patients’ recovery. The robotic arm and hand use digital algorithms to detect motions that patients cannot execute during therapy thus improving their performance per hour than they would have if worked with a physical therapist alone thus allowing speedy recover[4].Robots can also perform tests, x-rays, CT scans, data entry, and other tasks faster and more accurately. Cardiology and Radiology are two fields where the amount of data to analyze is huge and time-consuming. Future cardiologists and radiologists should look only at the most critical cases in which human monitoring is useful [5]. Caption Guidance, which is an Artificial Intelligence guided ultrasound platform or software capable of instructing clinicians on obtaining a clearer picture of the heart in motion. It will be used for capturing echocardiographic images of the patient’s heart without special training, spotting high-quality 2D heart images, and automatically recording video clips for later analysis, while calculating heart function measures thus improving the diagnosis of heart diseases [4].
Fig: AI tools in health care
Conclusion: Artificial Intelligence will surely improve the healthcare industry, from predictive medical care and more accurate diagnosis to motivating the patients to take care of their health. It will certainly continue enhancing the patient’s experience and healthcare expertise in general. The use of Artificial Intelligence is predicted to decrease medical costs as there will be more accuracy in diagnosis and better predictions in the treatment plan as well as more prevention of disease. It will not replace healthcare workers but instead allow them to spend more time for the bedside care of their patients, resulting in the greater outcomes for all.
References:
[1]Chan, Y., Chen, Y., Pham, T., Chang, W., & Hsieh, M. (2018, July 15). Artificial Intelligence in Medical Applications. Retrieved September 13, 2020, from https://www.hindawi.com/journals/jhe/2018/4827875/
[2]Says:, A., Says:, D., Says:, J., Says:, T., Says:, C., Says:, B., . . . *, N. (2019, June 19). Artificial Intelligence in Medicine: Applications, implications, and limitations. Retrieved September 13, 2020, from http://sitn.hms.harvard.edu/flash/2019/artificial-intelligence-in-medicine-applications-implications-and-limitations/
[3]Staff, S. (2018, November 06). Artificial intelligence predicts Alzheimer’s years before diagnosis. Retrieved September 13, 2020, from https://medicalxpress.com/news/2018-11-artificial-intelligence-alzheimer-years-diagnosis.amp
[4]Swetha. (2019, November 28). 10 Common Applications of Artificial Intelligence in Health Care. Retrieved September 13, 2020, from https://medium.com/artificial-intelligence-usm-systems/10-common-applications-of-artificial-intelligence-in-health-care-9d34ccccda5c
[5]Micah Castelo Micah Castelo is a web editor for EdTech: Focus on K-12 and a regular contributor for HealthTech. Her experience includes education and community news coverage for the Syracuse Post-Standard and international news reporting. (2019, May 01). The Future of Artificial Intelligence in Healthcare. Retrieved September 13, 2020, from https://healthtechmagazine.net/article/2020/02/future-artificial-intelligence-healthcare
Pratiksha Baliga, Youth Medical Journal 2020
Youth Medical Journal 